Demand sensing after the forecasting reset

Why it matters

Volatile demand in recent years exposed the limits of forecasts built mainly on historical sales. Retailers are adding near-term signals, such as weather, promotions, local events and online behaviour, to adjust plans in days instead of months.

Demand sensing works best when it is linked to decisions that can change quickly, such as store replenishment and promotion allocation. Where supply cannot respond within the forecast horizon, better short-term signals add less value.

What's changing

• Shorter planning cycles are replacing fixed monthly forecasts in many categories.
• Machine-learning models combine internal and external signals at store and product level.
• Planners spend more time on exceptions and less on routine adjustments.

How leaders are responding

• Piloting demand sensing in categories where volatility and waste are highest.
• Connecting forecasts directly to replenishment and allocation decisions.
• Tracking forecast accuracy and on-shelf availability together.

What good looks like

• Forecasts that blend sales history with live signals such as weather, events and search
• Planners who manage exceptions rather than adjust every number
• Forecast accuracy measured at the level decisions are made

What to watch

• Use of weather, events and search data in daily forecasts.
• Integration of forecasting with pricing and markdown decisions.
• Shared forecasts with suppliers to shorten lead times.

Where to start